bagasshw/whisper-large-v2-jv-full

This model is a fine-tuned version of openai/whisper-large-v2 on the jv_id_asr_split dataset. It achieves the following results on the evaluation set:

• Loss: 0.0848
• Wer: 7.1124

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 60000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2504	0.1351	5000	0.2563	19.9661
0.1976	0.2702	10000	0.2127	17.9671
0.1667	0.4052	15000	0.1803	14.7502
0.1599	0.5403	20000	0.1591	13.4658
0.1453	0.6754	25000	0.1421	12.3080
0.1273	0.8105	30000	0.1297	11.3644
0.1152	0.9456	35000	0.1169	10.6327
0.0524	1.0807	40000	0.1083	9.5124
0.053	1.2158	45000	0.1016	8.8378
0.0438	1.3508	50000	0.0956	8.3288
0.0384	1.4859	55000	0.0880	7.4525
0.0367	1.6210	60000	0.0848	7.1124

Framework versions

• Transformers 4.50.0.dev0
• Pytorch 2.7.0+cu128
• Datasets 2.18.0
• Tokenizers 0.21.1